FISHERY BULLETIN: VOL. 87. NO. 3, 1989 



interactions were addressed. Variation in preda- 

 tor performance was examined by calculating 

 the percentage of predator en-or occurring in 

 five groups of both types of predator feeding on 

 adult Artemia. Since Artemia show no avoid- 

 ance responses, a predator error was recorded 

 when a predator simply missed the prey. Per- 

 centage of predator error was calculated for each 

 predator group and among groups. In addition, 

 potential stress or mortahty of larvae due to 

 handling was examined. For each larval size 

 class, four replicates (trials) of five larvae each 

 were transferred into 76 L tanks containing no 

 predators. Mortality observed in control tanks 

 was used to adjust for handling-induced mortal- 

 ity or increased vulnerability of larvae in preda- 

 tion trials. 



Analysis of Larval Sensory System 

 Development 



Subsets of larvae from each size class were 

 removed from culture tanks for analyses of 

 larval visual and mechanoreceptive systems. 

 Groups of 6-8 larvae from each size class were 

 fixed in 5% phosphate-buffered formahn for his- 

 tological analysis of organogenesis. Larvae were 

 dehydrated, cleared, embedded in paraplast, 

 and serially sectioned at 5 (xm transversely and 

 sagittally. Sections were stained with Harris' 

 haemotoxylin, counterstained with eosin, and 

 viewed under light microscopy at 80-1250 mag- 

 nification. Ontogenetic development of the lens, 

 retina, and optic tectum were examined and 

 swimbladder inflation was noted. Size-specific 

 visual acuity was calculated from retinal sections 

 using the formula: sin a = elf, where a is the 

 minimum separable angle, c is the distance be- 

 tween the centers of adjacent cones, and/is the 

 focal length of the lens (Neave 1984). Cones were 

 measured as numbers per 100 ixm length of 

 retina (d), thus the reciprocal 10 d gives cone 

 separation in mm. The expression wao multiplied 

 by 1.11 to adjust for approximate 10% shrinkage 

 during processing, and the focal length of the 

 lens was calculated by multiplying its radius (r) 

 by 2.55 (Matthiessen's ratio; Matthiessen 1880), 

 giving: sin a = 0.0435/dr. 



Separate groups of 6-8 larvae from each size 

 class were examined for development of free 

 neuromast organs. These larvae were anaesthe- 

 tized with MS-222 and immersed in a bath of the 

 vital stain Janus Green (0.05% Janus Green 

 made up with 50% seawater) (Blaxter et al. 

 1983). Larvae were immersed for 20-30 minutes 



and then removed for examination of number 

 and location of free neuromasts under light 

 microscopy. 



Histological sections also were examined to 

 determine the timing of swimbladder inflation in 

 larvae. 



Data Analysis 



For each predation trial, predation rate (per- 

 cent killed in 10 minutes) was calculated from the 

 equation: A = m + n - mn, where A = total 

 mortality rate, m = control mortality rate, n = 

 experimental predation rate, and inn = the in- 

 teraction effect which estimates the proportion 

 of larvae consumed but which would have died 

 anyway from handling stress or other causes 

 (Ricker 1975). Percentage of larvae responding 

 and escaping were considered total survival 

 rates (S) and were calculated by first estimating 

 total conditional mortality (A) (proportion "not 

 responding" or "not escaping") and then calculat- 

 ing the difference as S = 1 - A (Ricker 1975). 



Lens diameter, visual acuity, and thickness of 

 the optic tectum were calculated for each larval 

 size class and developmental stage. Develop- 

 mental ontogeny of the retina, optic tectum, and 

 swimbladder were described. The mean number 

 of neuromast organs also was calculated, and 

 composite maps were developed of the patterns 

 of neuromast organ formation. 



Statistical analyses of data were performed 

 using SAS (SAS 1982) statistical programs. 



RESULTS 



Predator/Prey Interactions 



Potential Biases 



Preliminary determinations of predator error 

 by both predator types indicated that predator 

 performance would not bias results. Mean preda- 

 tor error per trial for northern anchovy adults 

 feeding on AHemla was 3.3% (range 2.1-5.1%) 

 while eiTor rate for juvenile white seabass was 

 2.7% (range 0.8-6.1%). There were no signifi- 

 cant differences in mean error rate among preda- 

 tor groups, within trials, or between species 

 (ANOVA, P>0.10). 



Larval mortality in control tanks (no preda- 

 tors) was low, ranging from 0.0 to 12.5% mean 

 values for any larval size group. Experimental 

 predation and escape rates were adjusted by the 

 control rates. 



540 



